> As far as I know, fres has never been pipelined. frsqrte and vrefp have been
> pipelined and have latency similar to multiply.
fres is pipelined on the IBM G3 (2 cycle throughput) and the G5 (1 cycle).
>You can use the frsqrte to do a pipelined divide with a bit of ingenuity,
You need to square the frsqrte result, losing a little more than a bit of
accuracy.
> but since it doesn't accept negative arguments, it is a bit more work.
fabs the argument and multiply the result by __fsel (a, 1.0, -1.0), where a
is the original argument.
> I am not aware of a available test to determine which type of G3 that you
have.
On Apple hardware, IBM G3 chips were used for G3 iMacs and notebooks running
above 700 MHz (which accounts for fairly few models). For a rigorous check,
you'd have to check the PVR register, or use Gestalt (though I seem to
recall Gestalt giving incorrect results in this case).
_______________________________________________
Do not post admin requests to the list. They will be ignored.
PerfOptimization-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/perfoptimization-dev/email@hidden
This email sent to email@hidden
